基于互信息的SWMM模型参数全局敏感性分析

Analysis of SWMM Parameters Global Sensitivity Based On Mutual Information

采用互信息方法对SWMM模型输入参数与输出结果之间的非线性关系进行了全局敏感性分析。以山东大学千佛山校区降雨径流模拟为例,采用拉丁超立方体抽样,研究了SWMM模型模型参数对输出结果的影响程度。结果表明:峰值流量最敏感的参数为透水区曼宁糙率系数;峰现时间最敏感的参数为管道曼宁糙率系数,其次为最小渗透率;总产流最敏感的参数为最小渗透率。

Mutual information method was used to conduct the overall sensitivity analysis on the complex non-linear relationship between the input parameters and outputs of Storm Water Management Model （SWMM） . Taking rainfall-runoff simulation in an area of Shandong Univer-sity as an example,the Latin Hypercube Sampling （LHS） method was used to sample values for input parameters of SWMM model,and the influence of input parameters to the model results was studied. The results show that the most sensitive parameter for peak discharge is per^me- able zone Manning value and the most sensitive parameter for peak time is pipeline Manning value, and then the minimum permeability; the most sensitive parameter of total output is the minimum permeability.